Physical Properties of the Mars Exploration Rover Landing Sites As Inferred from Miniв•'Tesв•ﬁderived Thermal Inertia

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 111, E02S21, doi:10.1029/2005JE002583, 2006

Physical properties of the Mars Exploration Rover landing sites as inferred from Mini-TES–derived thermal inertia Robin L. Fergason,1 Philip R. Christensen,1 James F. Bell III,2 Matthew P. Golombek,3 Kenneth E. Herkenhoff,4 and Hugh H. Kieffer5 Received 1 September 2005; revised 28 October 2005; accepted 8 November 2005; published 11 February 2006.

[1] The Miniature Thermal Emission Spectrometer (Mini-TES) on board the two Mars Exploration Rovers provides the first opportunity to observe thermal properties from the Martian surface, relate these properties to orbital data, and perform soil conductivity experiments under Martian conditions. The thermal inertias of soils, bedforms, and rock at each landing site were derived to quantify the physical properties of these features and understand geologic processes occurring at these localities. The thermal inertia for the Gusev plains rock target Bonneville Beacon (1200 J m2 K1 s1/2) is consistent with a dense, basaltic rock, but the rocks at the Columbia Hills have a lower thermal inertia (620 J m2 K1 s1/2), suggesting that they have a volcaniclasic origin. Bedforms on the floors of craters at both landing sites have thermal inertias of 200 J m2 K1 s1/2, consistent with a particle diameter of 160 mm. This diameter is comparable to the most easily moved grain size in the current atmosphere on Mars, suggesting that these bedforms may have formed under current atmospheric conditions. Along the Meridiani plains, the thermal inertia is lower than that derived from TES and Thermal Emission Imaging System (THEMIS) orbital data. This discrepancy is not well understood. Mini-TES–derived thermal inertias at Gusev along a 2.5 km traverse follow trends in thermal inertia measured from orbit with TES and THEMIS. However, along the traverse, there are variability and mixing of particle sizes that are not resolved in the orbital thermal inertia data due to meter-scale processes that are not identifiable at larger scales. Citation: Fergason, R. L., P. R. Christensen, J. F. Bell III, M. P. Golombek, K. E. Herkenhoff, and H. H. Kieffer (2006), Physical properties of the Mars Exploration Rover landing sites as inferred from Mini-TES–derived thermal inertia, J. Geophys. Res., 111, E02S21, doi:10.1029/2005JE002583.

1. Introduction interpretations made at the MER sites to other regions of the planet. [2] An important science goal of the Mars Exploration [3] The Miniature Thermal Emission Spectrometer Rover (MER) mission is to identify and understand current (Mini-TES) instruments on board both rovers provide processes acting on the Martian surface and how they have the first opportunity to (1) observe a surface’s diurnal shaped the morphology in recent history. Thermal inertia temperature variation and acquire Microscopic Imager is an important tool for understanding recent processes (MI) and Panoramic Camera (Pancam) imagery of that because it provides a means to quantify surface physical same surface from close proximity; (2) derive particle characteristics, such as particle size, clast abundance, and sizes from thermal inertia and compare them to grain size rock porosity or degree of weathering. This measurement distributions measured directly with the MI to validate also allows for quantitative comparisons between the laboratory conductivity measurements of grain size frac- MER sites. In addition, calculating thermal inertia at the tions at Martian atmospheric pressures; (3) test our ability MER sites is important for improving the interpretation of to adequately model the reflection, emission, and absorp- orbitally derived thermal inertias, increasing the confi- tion into the subsurface of solar and thermal energy and dence in the orbital measurements, and extending the predict the resulting surface temperature; and (4) validate orbital temperature and thermal inertia data. The Mini-TES 1 Department of Geological Sciences, Arizona State University, Tempe, experiment also provides the potential to better understand Arizona, USA. 2Department of Astronomy, Space Science Building, Cornell Uni- nonideal physical properties that can affect modeled tem- versity, Ithaca, New York, USA. perature results [e.g., Jakosky, 1979]. 3Jet Propulsion Laboratory, California Institute of Technology, [4] This work has two primary objectives. The first is to Pasadena, California, USA. 4 understand the interactions between nonideal surfaces and Astrogeology Team, U.S. Geological Survey, Flagstaff, Arizona, USA. measured temperatures, which will help determine if Martian 5Celestial Reasonings, Carson City, Nevada, USA. surface and subsurface properties are being adequately Copyright 2006 by the American Geophysical Union. represented in the current thermal models, and will improve 0148-0227/06/2005JE002583 the understanding of orbital data. The second goal is to

E02S21 1of18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21 accurately characterize the physical properties of surface molecules move between grain walls essentially unimpeded materials present at the MER landing sites (e.g., bedforms, by the presence of other gas molecules, resulting in an particle size distributions, abundance of rock fragments) and inefficient transfer of heat. Thus decreasing the particle size explore what these properties imply regarding recent pro- increases the number of grain-to-grain contacts per unit cesses that have affected the surface. length, and decreases both solid and gas conduction [Jakosky, 1986; Presley and Christensen, 1997a]. As a 1.1. Instrument Background result the thermal inertia is strongly controlled by the [5] The primary data sets utilized in this study are Mini- effective particle size, which describes the mean particle TES–derived surface temperatures (used to calculate ther- size of the upper few cm of surface material if that surface mal inertia values), the estimated bolometric albedos and was composed of unconsolidated spheres [Kieffer et al., images for surface morphology and context derived from 1973]. Surfaces dominated by fine-grained material have a the Panoramic Camera (Pancam) [Bell et al., 2003, 2004a, lower thermal inertia, and surfaces consisting of rock, 2004b, 2006] and Microscopic Imager (MI) [Herkenhoff et duricrust, unconsolidated sand, or any mixture of these al., 2003, 2004a, 2004b] images for measuring particle-size materials have a higher thermal inertia. Although thermal distributions. The Mini-TES [Christensen et al., 2003] is a inertia does not supply a unique solution to the particle size Fourier transform interferometer/spectrometer with spectral distribution of surface materials [e.g., Christensen, 1982], it 1 range of 339 to 1977 cm (5 to 29 mm). The instrument has does provide significant insight into the physical nature of two field of view (FOV) options with a spatial resolution of the surface and is uniquely related to an effective particle 6.9 and 17.5 mrad. The 17.5 mrad FOV was designed for size [Kieffer et al., 1973]. observing targets in the near field, and was utilized most often during the mission. 2. Methods [6] The Pancam [Bell et al., 2003] is a digital imaging system utilizing two 1024 1024 pixel frame transfer [9] Preliminary thermal inertia results at the Spirit and charge-coupled device (CCD) detector cameras with a Opportunity landing sites were reported by Christensen et 30-cm stereo separation and 0.27 mrad per pixel reso- al. [2004a] and Golombek et al. [2005]. Thermal inertia lution. Each camera has an eight-position filter wheel values reported here differ from those initial results due to (16 filters total, 13 of which are typically used for surface improvements in the Mini-TES instrument calibration (P. R. observations), and includes a spectral range of 436– Christensen et al., Calibration of a field-deployed infrared 1109 nm. spectrometer operating in an extreme, dusty environment: [7] The MI [Herkenhoff et al., 2003] is mounted on the The Mars Exploration Rover Mini-TES, manuscript in Instrument Deployment Device (IDD) robotic arm, and is preparation, 2006; hereinafter referred to as Christensen et a fixed-focus camera utilizing a 1024 1024 CCD. This al., manuscript in preparation, 2006), the inclusion of camera acquires images over a spectral range of 400– Pancam dust opacity measurements as model input param- 700 nm at a spatial resolution of 31 mm per pixel at best eters, varying both albedo and thermal inertia to fit diurnal focus [Herkenhoff et al., 2003]. Single grain sizes can temperature curves, and the use of full diurnal temperature be identified with confidence when they are larger than measurements rather than nighttime data only. 3 pixels, and this results in the ability to accurately measure grain diameters larger than 100 mm[Herkenhoff et 2.1. Modeling Thermal Inertia al., 2004a]. [10] Two techniques were developed in this study to calculate thermal inertia values: (1) fitting a full diurnal 1.2. Thermal Inertia Background curve when diurnal temperature measurements were avail- [8] Thermal inertia is controlled by the physical proper- able or (2) interpolation from a look-up table when a single ties of the upper few centimeters of the surface, and temperature measurement was used. is independent of time of day, season, and latitude. It is 2.1.1. Diurnal Temperature Technique 1/2 defined as I =(rkc) , where r is the bulk density, k is the [11] To calculate thermal inertia when diurnal tempera- thermal conductivity, and c is the specific heat, and is a ture measurements were available, we used Mini-TES– measure of the resistance of a material to changes in calibrated radiance and a thermal model similar to Kieffer temperature. Units are J m2 K1 s1/2 throughout this et al. [1977]. Model-derived diurnal temperature curves work (divide by a factor of 41.86 to convert to Viking-era are first calculated using the latitude, season, local time, 103 cal cm2 K1 s1/2 units). Under Martian conditions, elevation, and atmospheric dust opacity appropriate for the thermal inertia is primarily a function of the bulk conduc- observation. The latitude, season, and local time are tivity of the surface because the density and specific heat determined from spacecraft ephemeris, and the elevation of geologic materials vary by about a factor of 4, while is derived from a MOLA [Smith et al., 2001] elevation thermal conductivity varies by up to 3 orders of magnitude map binned at 8 pixels per degree (1.6 km and 1.5 km [Wechsler and Glaser, 1965; Neugebauer et al., 1971; below datum at Gusev crater and Meridiani Planum, Kieffer et al., 1973]. The bulk conductivity is a function respectively). The dust opacity is determined from Pancam of both the solid and gas conductivity, which is largely measurements of the visible dust opacity (t) taken on the controlled by particle size and the relationship between same day [Lemmon et al., 2004]. pore size and the molecular mean free path of the gas. [12] The Mini-TES data are obtained at multiple times of At Martian pressures the gas molecule mean free path is day to provide good diurnal coverage, and for a minimum 5 mm. When the pore size is comparable in size or smaller of 15 minutes per observation to increase the signal-to-noise than the gas mean free path (relatively small grains), gas ratio. Each spectrum acquired is converted to a surface

2of18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21 target temperature by fitting a Plank function to the cali- also emitted from the surface, causing some additional brated radiance at each wavelength. The warmest brightness energy to reflect off the atmosphere back onto the surface. temperature between 500 and 1700 cm1,selectedto In addition, emission from objects, such as rocks, in close avoid wavelength regions where the signal is low, is used proximity to the target of interest must also be considered. to approximate the surface kinetic temperature, and defines The thermal model attempts to account for all of these the target temperature [Christensen et al., 2004b]. effects, and areas of largest uncertainty include the [13] Data obtained at night are important for accurately amount of energy reflected or conducted into the subsur- modeling the thermal inertia because the effects of Sun- face and the amount of downwelling or backscatter heated slopes and albedo on surface temperature are radiance from the atmosphere or nearby objects. By minimized, and the thermal contrast due to differences allowing albedo to be a free parameter, we are directly in particle size is at a maximum [e.g., Kieffer et al., 1977; correcting for errors in albedo and indirectly accounting Palluconi and Kieffer, 1981; Christensen, 1982]. The cool for the downwelling radiance, and the amount of heat nighttime surface does have a lower thermal flux than energy absorbed by the surface is more accurately esti- the daytime surface, which causes the uncertainty in the mated [e.g., Palluconi and Kieffer, 1981; Hayashi et al., measured surface temperature to increase. However, the 1995]. Allowing albedo to be a free parameter, rather than reduced contribution of albedo and slopes on surface using Pancam albedo values, typically changes the best-fit temperature at night compensates for this effect, allowing thermal inertia by less than 10%, but reduces the RMS a more reliable derivation of thermal inertia. We define residuals. night as between 21 and 6 H (24 H equals one Martian 2.1.2. Single Temperature Technique day and 0 H is local midnight) because the diurnal curves [16] Along the traverse at Gusev and when sampling the are roughly parallel and linear at this time. Thermal rocks at the Colombia Hills, Mini-TES measurements were inertias were primarily derived for surfaces with nighttime acquired at a single time of day, requiring a different method measurements to ensure the most accurate determination of for calculating thermal inertias. To determine the thermal thermal inertia. The majority of night observations oc- inertia with a single measurement, this temperature was curred during the primary mission (first 90 sols; sol is interpolated using a seven-dimensional look-up table. To defined as the mean length of a Martian day, 24.7 hours construct this look-up table, surface kinetic temperatures [Kieffer et al., 1992]) because the reduction in power as were calculated for a range of albedo, thermal inertia, dust settled on the solar panels [Arvidson et al., 2004a, latitude, local time, season, elevation, and dust opacity 2004b] made acquiring night measurements more difficult values using the same thermal model as the diurnal tem- later in the mission. perature technique. For the interpolation, latitude, local [14] To determine the thermal inertia, albedo and thermal time, season, elevation, and dust opacity values appropriate inertia are varied as model input parameters until a best fit for the observation were used. An albedo of 0.22 for the between model-derived diurnal temperature curves and Gusev Plains traverse and 0.15 for rock was assumed for calculated target temperatures of the observed scene are each observation, which corresponds to 0.03 (the average determined using a root-mean square minimization. Some difference between the thermally derived albedo and the previous thermal inertia data sets, such as TES, used a Pancam albedo) less than the average Pancam albedo for single-point method to calculate thermal inertia values these surface types. [Jakosky et al., 2000; Mellon et al., 2000]. The technique 2.1.3. Model used here differs because diurnal temperature measurements [17] The thermal model used in this study is derived are available, and fitting the full diurnal curve is a more from the Viking Infrared Thermal Mapper (IRTM) thermal accurate method for calculating the thermal inertia. This model [Kieffer et al., 1977, Appendix 1]; the primary method is similar to that used with Viking data when diurnal modification has been replacement of a constant atmo- temperature measurements were available [e.g., Palluconi spheric thermal radiation with a one-layer, radiatively and Kieffer, 1981]. The thermally derived albedo was also coupled atmosphere of the appropriate thermal capacity. compared to Pancam estimated bolometric albedo (here The atmosphere is spectrally gray at solar wavelengths, after referred to as Pancam albedo). The Pancam albedo and direct and diffuse illuminations are computed using a was estimated following the method of Bell et al. [2006] 2-stream delta-Eddington model. Thermal radiation is and Reid et al. [1999], where R* is calculated by dividing assumed to be gray and isotropic, with a fixed ratio of the 750 nm filter IOF image by the cosine of the solar infrared-to-visible opacity. An explicit forward finite dif- incidence angle at the time of the observation and is an ference scheme calculates surface and subsurface temper- approximation of the Lambert albedo within each Pancam atures by solving the heat diffusion equation while band pass [Bell et al., 2006]. The absolute radiance cali- satisfying a surface boundary condition including direct bration and the relative reflectance of the Pancam instru- and diffuse insolation, upward emission and downwelling ment (i.e., IOF) are estimated to be within 5–10% absolute thermal radiation, and the latent heat of CO2 if its accuracy [Bell et al., 2004a]. saturation temperature is reached. This model assumes [15] The thermal model is a simplified version of a lateral uniformity and a Lambert surface reflection. Sub- complex system, which is used to predict the surface surface layers increase in thickness exponentially with temperature. In this system, solar and thermal energy from depth and are scaled to the diurnal skin depth. In this the Sun is filtered through the atmosphere, and some of work, the surface emissivity is assumed to be unity and the that energy reaches the surface and is either reflected back lower boundary is assumed to be insulating. This model into the atmosphere or is conducted into the subsurface; can also incorporate the effects of a radiatively coupled this division is controlled by the surface albedo. Energy is sloped surface at any azimuth. The effects of condensate

3of18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21 clouds, the latent heat of water ice, and three-dimensional average uncertainty of 10% during the day and 3.5% at blocks on the surface are not considered. night. 2.1.4. Thermal Inertia Derivation Uncertainties [22] There are also errors due to uncertainties in the [18] Different uncertainty determination methods were assumed visible/9-mm extinction opacity ratio. A constant implemented for each of the two thermal inertia derivation ratio of 2.0 was used in this work, but this ratio may vary techniques. For diurnal temperature measurements, the between 2.0 and 2.5 for atmospheric dust opacity thermal inertia errors resulting from model parameter uncer- conditions below 1.0 [Clancy et al., 1995], which includes tainties were determined using the root-mean square (RMS) all of the observations used in this study. Increasing the residual of the measured temperatures compared to the visible/9-mm extinction opacity ratio increases the modeled modeled diurnal temperature curve using the formula temperature at all times of day with a greater effect at night. RMS = (X2)1/2. Errors derived from the RMS are conser- Varying visible/9-mm extinction opacity ratio from 2.0 to 2.5 vative, as the rapid decrease of afternoon temperatures can changes the modeled temperature 0.5 K and 1.5 K produce a large difference in temperatures relative to the during the day and night respectively for low opacity true quality of the fit. Also thermally derived particles sizes conditions (t = 0.2) and 1 K and 3 K during the day agree with distributions measured directly with MI, further and night respectively for high opacity conditions (t = 0.6), indicating that the reported errors are conservative. producing 10% uncertainty. [19] Along the traverse at Gusev, and when sampling the rocks at the Colombia Hills, Mini-TES measurements were 2.2. Deriving Particle Sizes acquired at a single time of day, requiring a different method [23] Particle sizes derived from thermal inertia measure- for calculating thermal inertias and assessing their accuracy. ments are calculated using the technique described by The largest sources of error in this technique include the Presley and Christensen [1997a]. From laboratory measure- calibration of the instrument (7%), albedo estimation (17%, ments under Martian conditions, Presley and Christensen assuming a 0.01 uncertainty), atmospheric dust opacity [1997a] determined a linear relationship between the log of (9%, assuming a 0.02 uncertainty), and sloped surfaces. the thermal conductivity and the log of the particle diameter The highest slopes observed along the Gusev traverse were and derived parameters that relate conductivity to particle 3°, and can result in uncertainties up to 20% with an size using the following equation: average uncertainty of 10% during the day. These errors ÀÁ are believed to be uncorrelated, and therefore result in a root k ¼ CP0:6 d0:11 logðÞ P=K ; ð1Þ sum square (RSS; defined as S (x2)1/2) error of 12%. This is likely an underestimation, as there are additional uncertain- where C and K are the constants 0.0015 and 8.1 104 torr ties such as the effect of downwelling radiance and errors in (to convert to Pascal units, multiply 1 torr by 133.3 Pa), the visible/9-mm extinction opacity ratio that are difficult to respectively [Presley and Christensen, 1997a], P is atmo- quantify. However, the thermal inertia derived from Mini- spheric pressure in torr, and d is the particle diameter in mm. TES corresponds well with orbital data, establishing confi- This equation can be substituted into the definition of dence in these results. thermal inertia to obtain an equation that relates particle size [20] The absolute calibration of the Mini-TES instrument to thermal inertia: determines the accuracy with which the kinetic surface temperature can be approximated, and is a function of both ÀÁ 2 0:6 1=0:11* logðÞ P=K the precision and the accuracy of the instrument. The Mini- d ¼ I =rcCP ; ð2Þ TES–derived target temperature has a precision and accuracy of ±0.5 K and ±2 K, respectively for nighttime where rc is assumed to be 1 106 Jm3 K[Neugebauer et measurements and ±0.1 K and ±0.5 K for daytime measure- al., 1971]. This equation was used to derive an effective ments for the 17.5 mrad FOV [Christensen et al., 2004b], particle size [Kieffer et al., 1973] from thermal inertia which corresponds to a 7% uncertainty in thermal inertia assuming an atmospheric pressure of 600 Pascals (Mars calculations. For further details on the calibration of the surface pressure at 1.5 km and Ls =0° [Smith and Zuber, Mini-TES instrument, see Christensen et al. (manuscript in 1998]). Equation (1) is valid for thermal inertias less than preparation, 2006). 350. Thermal inertias larger than this value are more [21] The sensitivity of the thermal model to variations in difficult to interpret, but this relationship provides a input parameters (e.g., albedo, elevation, and atmospheric reasonable estimate. Errors in deriving particle sizes with dust opacity), and how these sensitivities propagate into this method are expected to be less than 10–15% [Presley thermal inertia uncertainties is important for understanding and Christensen, 1997a]. These thermally derived particle the sources of error in each observation. These sensitivities sizes are compared to particle sizes measured directly in MI were investigated by varying each key parameter while all images, where available, to validate the accuracy of this other parameters were held constant, allowing the determi- technique. nation of the partial derivative of the individual parameter [24] There are several caveats to calculating particle sizes with respect to thermal inertia for each model input. For the from thermophysical properties. This technique requires diurnal temperature technique used in this study, uncertain- that the surface consists of unconsolidated sediments down ties in atmospheric dust opacity and slope measurements are to at least a diurnal thermal skin depth [e.g., Jakosky, 1986], the largest sources of error. An uncertainty in opacity of and that particles are loosely packed and of a single, 0.02 [Lemmon et al., 2004] results in a 9% error during the spherical grain size. Packing of grains, mixtures of particles day and 0.75% at night in the resulting thermal inertia. where pore spaces between larger grains are filled by Errors in slope measurements of 3°,andresultsinan smaller ones, and nonspherical grains typically increase

4of18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21

Table 1. Thermal Inertia, Uncertainties, Thermally Derived Albedo, and Pancam Estimated Bolometric Albedo for Each Mini-TES Diurnal Observation Pancam Estimated Thermal Inertia Temperature RMS D Inertia RSS Thermally Derived Albedo Bolometric Albedo Gusev Middle Ground 150 1.950 28 (18.7%) 0.18 0.27 Bonneville Upper Dunes 200 2.467 40 (20%) 0.16 0.18 Bonneville Lower Dunes 160 2.286 35 (21.9%) 0.16 0.23 Saber 250 2.896 42 (16.8%) 0.27 0.29 Beacon 1200 1.707 295 (21.1%) 0.11 0.18

Meridiani Plains 100 6.340 71 (71%) 0.12 0.12 150 5.912 72 (48%) Endurance Dunes 200 2.037 28 (14%) 0.09 0.12 Endurance Sand 100 1.589 23 (23%) 0.14 0.13 the conductivity [Fountain and West, 1970; Wechsler et al., of dust particles suspended in the atmosphere (2.5 mm 1972; Presley and Christensen, 1997b]. Many of these [Pollack et al., 1979; Toon et al., 1977]), and the Mini-TES conditions occur in natural surfaces, and future laboratory emissivity spectra of the Laguna trench material contains measurements addressing heat conduction in crusts, particle features that are consistent with basaltic material [Wang et size mixtures, and nonspherical grains are needed in order to extend thermal inertia observations from simple unimodal granular material to more complex surfaces.

3. Results and Discussion

[25] The results of this study are separated into two sections: (1) thermal modeling results of major Martian surface types identified from orbit, which are dust, sand, and rock, and (2) specific examples from both Gusev crater and Meridiani Planum. A summary of thermal inertias, uncertainties, thermally derived albedo values, and Pancam albedo values are provided in Table 1. Because the accuracy of the thermal inertia calculation differs for day and night measurements, the RSS (defined as S (x2/n)1/2)of uncertainty for each observation is reported.

3.1. Surface Types 3.1.1. Dust [26] Some of the finest-grained material observed at the MER landing sites is located in shallow sediment-filled depressions in Gusev crater. These features, named hollows, are interpreted as having been formed by impacts and filled with aeolian material [Grant et al., 2004; Golombek et al., 2006]. One example is Middle Ground (Figure 1a), which is roughly equidistant from the landing site named Columbia Memorial Station (CMS) [Squyres et al., 2004a] and Bonne- ville crater, and was observed on MER-A sols 55–57. The modeled thermal inertia of this material is 150 (Figure 1b), which implies 45 mm diameter particles (silt). The thermally derived albedo is 0.18, and the Pancam albedo is 0.27. A similar hollow, Laguna (40 m from Middle Ground), was trenched with the rover wheel to expose cohesive material to a depth of 6–7 cm [Squyres et al., 2004a]. MI images of the Laguna trench indicate that grains are sometimes clumped together, but individual Figure 1. Gusev crater hollow. (a) Middle Ground particles are too small to measure with the MI instrument Hollow (Pancam approximate true color image (less than 100 mm) (Figure 2). This observation is consistent 2P131347008FFL1155P2584, sol 056). For scale, the with the thermal inertia values and particle sizes derived Pancam calibration target in the lower center of the image from Mini-TES. is 20 cm. (b) Plot of the modeled diurnal temperature [27] The particle sizes, both implied by the thermal inertia curve (using an opacity of 0.76 and Ls of 357.5°) and and measured using MI images, are higher than the diameter Mini-TES target temperatures; RMS error is 2.0 K.

5of18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21

al., 2006] greater than 60 mm [e.g., Ruff and Christensen, 2002]. Thus it is unlikely that the material inside the hollows originated from airfall dust alone, and it is instead probable that aeolian transportation of fine material from the plains and the deposition of airfall dust is filling these hollows simultaneously. There are also few rock fragments inside the hollows, in contrast to the surrounding plains, and their absence is potentially the cause for the reduced thermal inertia in the hollows, relative to the surrounding region. These observations support the hypothesis of Grant et al. [2004] who proposed that regionally derived fines, possibly resulting from impacts, have filled the hollows on the Gusev plains. 3.1.2. Sand [28] The bedforms on the floor of Endurance crater at Meridiani Planum was observed on MER-B sols 211–215 (Figure 3). These bedforms have sharp crests and a Pancam albedo 0.12, suggesting a relatively dust-free surface. The bedform crests have many different orientations, providing evidence for a complicated wind regime that is likely due to the crater topography. These bedforms have an overall net zero slope, but the viewing geometry restricted Mini-TES to observe primarily west-facing Figure 2. Subsurface of Gusev crater hollow. MI image slopes and some interbedform material. This morphology 2M130618323FFL09BVP2955 (sol 048) of the subsurface was approximated in the determination of thermal inertia of Laguna Hollow. The inability to obtain a focused image by incorporating equal components of a horizontal surface suggests particles below the instrument resolution (less than and a west-facing slope at an angle of 15°, causing the 100 mm). bedform to receive more heat in the afternoon (Figure 4a).

Figure 3. Meridiani Planum Endurance crater bedforms. Pancam approximate true color mosaic of Endurance bedforms, including the approximate location of the Mini-TES observations.

6of18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21

suggesting a lack of coating or dust mantle. Bonneville Beacon is 75 cm in length and is larger than the diurnal skin depth of solid rock (15 cm [Christensen, 1986]); no temperature gradient along the rock was observed. Since Mini-TES is observing the outside surface of the rock and Bonneville Beacon is only 5 thermal skin depths in length, boundary effects may be causing the warmer rock surface thermal inertia to be lower than that of the rock interior because warmer temperatures are consistent with lower thermal inertias during the day. The slope and azimuth angles of the rock faces viewed by Mini-TES were estimated using Pancam imagery. Diurnal temperature curves modeled for a slope angle of 60° at azimuth angles of 45° and 160° were added in equal proportions to best represent the orientation of the rock target causing mid-day and afternoon temperatures to be modeled 10°K and 3°K lower, respectively, than a horizontal surface (Figure 6a). Incorporating these slopes has an effect similar to that of increasing the thermal inertia, and therefore this thermal inertia estimate is likely a lower limit. The modeled thermal inertia after accounting for slopes is 1200 (Figure 6b), and the thermally derived albedo is 0.11. The thermal inertia of a basalt rock is expected to be 2200 (assuming r = 2800 kg m3,k= 2.1 J s1 m1 K1, c = 840 J K1 kg1 [Kahle, 1980]), which corresponds to a temperature difference of 2.5 K during the day and 4.5 K at night between these two thermal inertia values. The thermal inertia is that expected for a basaltic rock this size, and given the uncertainties in measuring surface temperature, is comparable to that of an infinite slab of basalt. The lower modeled thermal inertia may potentially be due to weathering, the presence of a Figure 4. Diurnal temperature curve of Meridiani Planum dust coating, micro-fractures within the rock [Robertson Endurance crater bedforms. (a) Plot of the modeled diurnal and Peck, 1974; Zimbelman, 1986], or the Mini-TES temperature curve for a flat-lying surface (RMS error is footprints being closer than 15 cm from the rock edge 3.3 K) and a sloped surface (RMS error is 2.0 K) causing warmer temperatures to be observed. approximating the morphology of Endurance bedforms, and Mini-TES target temperatures. (b) Plot of the 3.2. Surface Properties at the MER Landing Sites modeled diurnal temperature curve (using an opacity of 3.2.1. Spirit Landing Site in Gusev Crater 0.50 and Ls of 80°) and Mini-TES target temperatures; 3.2.1.1. Gusev Plains Soil RMS error is 2.0 K. [30] At Gusev, a campaign was implemented to system- atically monitor properties of the soil (following the con- The average modeled thermal inertia of the entire bedform vention of previous authors [e.g., Squyres et al., 2004a], soil is 200 (Figure 4b), corresponding to a particle diameter of is used here to denote any loose, unconsolidated materials 160 mm (fine sand), and the thermally derived albedo is that can be distinguished from rocks, bedrock, or strongly 0.09. The rover did not traverse to these bedforms cohesive sediments; no implication of the presence or because the area surrounding the bedforms was untraffi- absence of organic materials or living matter is intended) cable, and thus an MI measurement of the same surface along the traverse from Columbia Memorial Station (CMS) observed by the Mini-TES was not acquired. However, an to Bonneville crater and the Columbia Hills. This included, MI image of a patch of sand near the wall of the crater among other observations, measurements of the surface that is presumably the same material as the bedforms was radiance with Mini-TES to model thermal inertia and acquired. The average measured particle size of this acquisition of Pancam imagery of the same surface for bedform was 130 mm, which corresponds well with context. From Pancam imagery, these surfaces are com- the particle diameter derived from the Mini-TES thermal posed of a combination of fines, pebbles, cm-sized rock inertia value. fragments, and rocks (Figure 7). The results of this survey 3.1.3. Rock (Figure 8) indicate a consistently low thermal inertia (avg. [29] The rock informally named Bonneville Beacon at 175 ± 20, corresponding to particles of 90 mm, silt) Gusev crater (Figure 5) was observed on MER-A sol 47 from the CMS (Figure 7a) to the Bonneville crater ejecta, during the day only. The rover was also trenching in Laguna where the thermal inertia values increase sharply. The Hollow during this time, and power constraints for these observed increase in cm-sized rock fragments and higher operations were significant. This rock is darker (Pancam rock abundance in the ejecta [Golombek et al., 2005, albedo of 0.18) than rocks typically observed on the plains, 2006] suggests that this sharp increase in thermal inertia

7of18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21

Figure 5. Gusev crater rock Bonneville Beacon. Pancam approximate true color image (2P130443835FFL0900P2555, sol 046) of the rock target Bonneville Beacon and the approximate locations of the Mini-TES footprints that were used in this analysis. Bonneville Beacon is 0.75 m in length. does not indicate a change in soil properties, but instead larger than 10–15% [Presley and Christensen, 1997a]. reflects the increased number or size of clasts or rock in However, the range of thermal inertia values for rock the Mini-TES FOV. From Bonneville crater to the Colum- fragments (540 to 820) is a reasonable estimate, assuming bia Hills, the thermal inertia of the soil is higher (avg. that the relationship between conductivity and particle size 250 ± 30) than from CMS to Bonneville crater, and there established using laboratory data and extrapolated to rock is more variability in the thermal inertia measurements. follows a smooth curve [Kieffer et al., 1973]. Jakosky The greater variability is likely caused by changes in the [1986] argued that particle diameters of 1 mm to < a few amount of rock fragments and aeolian drift material in this cm have a constant thermal inertia of 420. Since the area. The traverse from Bonneville crater to the Columbia measured rock fragments ranged in size from 1.5 to 4 cm Hills also has an overall higher thermal inertia. This this constant value was assumed to be the thermal inertia for increase is probably due to the presence of more rock all rock fragments. From this thermal inertia, a temperature fragments as observed in Pancam images (Figures 7b, 7c, and then a surface radiance of the rock component was and 7d). derived. The percentage of rock in the Mini-TES FOV was [31] For scenes containing rock fragments and when estimated from Pancam imagery, and this percentage of nighttime temperature observations were acquired, the ther- calculated rock radiance was subtracted from the measured mal inertia of the soil alone was computed by removing the radiance of Mini-TES to obtain the radiance of the entire effects of the fragments. The average length of rock frag- target minus the radiance contribution from the rocks using ments in the Mini-TES FOV was measured using Pancam the following relationship: imagery, and two techniques were used to estimate their thermal inertia. The rock fragment thermal inertia was Rm ¼ as * Rs þ ðÞ1 as * Rr; ð3Þ estimated using equation (1) [Presley and Christensen, 1997a], and because these rock fragments have a thermal where Rm is the measured radiance with Mini-TES, as is the inertia greater than 350, the resulting uncertainties may be percentage of soil in the Mini-TES FOV, Rs is the radiance

8of18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21

result suggests that the soil properties do not change significantly along the traverse from CMS to Bonneville crater and that the increase in observed thermal inertia from orbit is due to additional rock abundance as the rover approached the crater ejecta material, in agreement with earlier results [Christensen et al., 2004a; Golombek et al., 2005, 2006; Moersch et al., 2005]. 3.2.1.2. Aeolian Bedforms [32] The most prominent bedform observed at the Gusev site is located in Bonneville crater (Figure 9). This aeolian material is prevalent on the crater floor and part of the south-facing wall. Rocks were observed poking through this aeolian sediment suggesting that this deposit is less than a meter or two thick [Grant et al., 2004]. Since the diurnal skin depth of medium sand is 7.5 cm [Edgett and Christensen, 1991] and there are no rocks observed within the Mini-TES FOV, the measured temperatures are likely determined by the bedform material and are not influenced by any potential coarser material on the crater floor. Two sections of this bedform were modeled: (1) an upper section that climbs the north crater wall (modeled slope of 11° [Grant et al., 2004]) and has a lower Pancam albedo (0.18) and (2) a lower portion that occurs on the crater floor and has a higher Pancam albedo (0.23). The modeled thermal inertia for the upper bedform section is 200 (Figure 10a), which corresponds to a particle diameter of 160 mm (fine sand). The lower bedform section has a thermal inertia of 160 (Figure 10b), suggesting 60 mm diameter particles (silt). Both areas have a thermally derived albedo of 0.16. The aeolian sediment on the crater floor and wall may be of the same origin, and a layer (1 cm) of dust may be causing the increased Pancam albedo and lower thermal inertia on the Figure 6. Diurnal temperature curve of Gusev crater rock crater floor. The coarser, darker grains of the upper Bonneville Beacon. (a) Plot of the modeled diurnal bedform section are consistent with bedforms that have temperature curve for a flat-lying surface (RMS error is been active more recently than the lower bedform section 12.0 K) and a sloped surface (RMS error is 1.7 K) because they presumably have less dust mantling their approximating the geometry of the rock Bonneville Beacon, surface. and Mini-TES target temperatures. (b) Plot of the modeled [33] The Saber and Serpent targets are aeolian bedforms diurnal temperature curve (using an opacity of 0.76 and Ls near the rim of Bonneville crater. The rover disturbed of 352.7°) and Mini-TES target temperatures; RMS error is Serpent using its wheel, which revealed an outer armor of 1.7 K. coarser grains that may be a lag deposit of windblown material that has been transported in traction by the of the soil, and Rr is the estimated radiance of the rock impact of smaller saltating grains [Greeley et al., 2004]. component. This soil radiance is converted to a brightness In addition, the interior material was composed of dust temperature, then a derived thermal inertia, to obtain an well mixed with sand, indicating that the sand grains are estimate of the thermal inertia of the soil. Errors in this not currently saltating [Greeley et al., 2006]. It is likely technique are primarily due to inaccuracies in measuring the that the Saber bedform has an armor and interior similar length of rock fragments (<5%), estimating the percentage to Serpent because of the similar albedo, temperature, of rock in the Mini-TES FOV (<5%), uncertainties in morphology, and proximity of these bedforms. The ther- calculating the thermal inertia (12%), and uncertainties in mal inertia of the undisturbed Saber bedform is 250, the Mini-TES instrument calibration (7%). Thermal inertia corresponding to a particle diameter of 415 mm (medium estimates for the minimum, maximum, and average size sand). This bedform has a thermally derived albedo of rock fragments at each location using equation (2) differed 0.27 and a Pancam albedo of 0.29. These results corre- by less than 10%, and using a rock fragment thermal inertia spond to MI images of the disturbed Serpent bedform of 420 [Jakosky, 1986], rather than 540 to 820, produced a (Figure 11a), indicating a trimodal distribution of particle thermal inertia of the soil 10 units lower. The estimated sizes of 1–2 mm, 250–500 mm and below 210 mm thermal inertia of the soil at the measured locations is [Herkenhoff et al., 2004a]. A layer of 1–2 mm particles 200 ± 20, corresponding to a particle diameter of 160 mm does not significantly affect the thermal inertia of this (fine sand). The thermal inertia of the soil was measured bedform. Although the 1–2 mm armor does increase the at MER-A sol 45 and sol 65 (200 m apart), and both thermal inertia, in this layered case the surface temperature measurements were within 50 units of one another. This is dominated by the fine-grained substrate (Figure 11b).

9of18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21

Figure 7. Type examples of Gusev plain surfaces. Pancam approximate true color images along the traverse: (a) 2P130264291FFL0700P2540 (site 7, sol 044), (b) 2P136034758FFL3300P2460 (site 33, sol 109), (c) 2P239052324FFL5500P2460 (site 55, sol 143), and (d) 2P143136938FFL7100P2460 (site 71, sol 189).

The armor is a small percentage of the total bedform, suggests that Bonneville Beacon is nonvesicular igneous and thus the thermal energy is conducted rapidly material, rather than a sedimentary or volcaniclasic rock, through the 1–2 mm layer, but is then impeded by and may be a portion of a basaltic lava flow that has been the smaller particles. This result also implies that there are exhumed from the subsurface by impact events. few 1–2 mm grain particles below the armor, as these [35] Rocks at the lower Columbia Hills have a different grains would dominate the observed bedform thermal chemical composition and physical appearance than rocks inertia [Presley and Christensen, 1997b]. Thus the ther- observed on the plains [Arvidson et al., 2006], but a high- mally derived particle size (415 mm) is consistent with quality diurnal temperature curve has not yet been acquired MI observations. in the Columbia Hills. So to assess the thermal inertia of 3.2.1.3. Rock and In-Place Bedrock rocks in this region, 11 rock targets with single Mini-TES [34] The rock informally named Bonneville Beacon at measurements were chosen on the basis of the following Gusev crater has a modeled thermal inertia of 1200, criteria: (1) the Mini-TES footprint was located completely and a thermally derived albedo of 0.11 (Figures 5 and 6). on the rock, (2) the rock size was more than 15 cm in The thermal inertia is that expected for a basaltic rock diameter (diurnal skin depth of rock [Christensen, 1986]), this size, as the thermal inertia of a solid, dense, infinite (3) no rover shadow was present, and (4) the observation slab of basaltic rock is 2200, and would have a was taken before 14 H because at later times of day the temperature within 2.5 K during the day and 4.5 K at diurnal curves converge resulting in all thermal inertia night of the temperature at Bonneville Beacon. This lower values having similar temperatures. The average thermal thermal inertia may be due to Mini-TES observing a warmer inertia of any rock selected was 620 (maximum was rock surface relative to the cooler interior, or from physical 1100 ± 130). This average thermal inertia value is lower characteristics such as weathering, the presence of a dust than the Bonneville Beacon thermal inertia on the Gusev coating, or micro-fracturing [Robertson and Peck, 1974; Plains (1200) resulting in an 5 K temperature difference Zimbelman, 1986] within the rock. This comparison during the day. This variation suggests that thermal inertia

10 of 18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21

Figure 8. Comparison of Mini-TES–derived surface and THEMIS-derived orbital thermal inertia values. Plot comparing the thermal inertia trends between THEMIS-derived thermal inertia from orbit and Mini-TES–derived thermal inertia of the surface. can discriminate real differences in the density or porosity orate the thermal inertia differences. A possible interpre- of rocks at different locations, and differentiate between tation is that the Gusev plains primarily consist of dense dense and friable, weathered rocks. On the basis of the basaltic rock, whereas the lower Columbia Hills is con- amount of energy necessary to grind rocks using the Rock sistent with friable volcaniclastic ash or impact debris Abrasion Tool (RAT), the rocks at the lower Columbia Hills [Arvidson et al., 2006]. are easier to grind (average energy per volume of rock 3.2.2. Opportunity Landing Site in Meridiani Planum removed of 15 J/mm3) than rocks on the Gusev plains 3.2.2.1. Meridiani Plains Soil 3 (average energy per volume of 52 J/mm ) (S. Gorevan et [36] The surface at Meridiani Planum is dominated by al., Analytical developments and results for rock specific low ripples (1 cm) spaced an average of 10 m apart, and grind energy from the Mars Exploration Rover Rock is covered with 0.6–6 mm diameter hematite spherules Abrasion Tool, manuscript in preparation, 2006; hereinaf- [Herkenhoff et al., 2004b], and a set of aeolian bedforms ter referred to as Gorevan et al., manuscript in prepara- interpreted as wind ripples [Squyres et al., 2004b; Bell et tion, 2006). Since the energy required relates directly to al., 2004b; Soderblom et al., 2004; Sullivan et al., rock density, among other properties, these results corrob- 2005]. The spherules are relatively free of adhering dust

Figure 9. Bedforms in Bonneville crater. The intracrater bedforms in Bonneville crater (Pancam approximate true color image 2P132398796FFL1800P2285, sol 068) and the locations of the Mini-TES footprints that were used in this analysis. For scale, Bonneville crater diameter is 210 m.

11 of 18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21

allow the surface to retain more heat in the afternoon, and this layered subsurface would not follow model predicted diurnal behavior. However, bedrock close enough to the surface to affect the thermal inertia is expected to outcrop along the traverse, which was rarely observed. Thus this scenario is not likely. In addition, the modeled thermal inertia is lower than what is implied by MI particle size measurements as the MI observed a bimodal particle size distribution of 0.6–6 mm diameter particles set in fine sand <125 mm[Herkenhoff et al., 2004b]. A material consisting of vesicular, less conductive grains, such as pumice, would lower the thermal inertia relative to the MI-measured particle size. This is also unlikely since the MI-measured particle sizes correspond well with orbital thermal inertia data [Fergason and Christensen, 2003; Christensen et al., 2004b]. In light of this discussion, there is not a clear cause at this time for the discrepancy between thermally derived particle sizes and particles

Figure 10. Diurnal temperature curves of bedforms in Bonneville crater. (a) Plot of the modeled diurnal temperature curve for the upper section of the bedform (using an opacity of 0.69, Ls of 4.6°, and slope of 11.0° at an azimuth of 180°) and Mini-TES target temperatures; RMS error is 2.5 K. (b) Plot of the modeled diurnal temperature curve for the lower section of the bedform (using an opacity of 0.69, Ls of 4.6°, and no slope) and Mini-TES target temperatures; RMS error is 2.3 K.

(Figure 12a), unlike the surface material at the Gusev landing site region, implying a different aeolian environment at these two localities. [37] The Meridiani Plains soil was observed during MER-B sols 88–91 and the thermal inertia is 100 to 150 (Figure 12b), corresponding to a particle diameter of 45 mm (silt). Both the thermally derived albedo and the Pancam albedo are 0.12. In this location, a single thermal inertia does not fit the entire diurnal curve. A thermal inertia of 100 fits the data well until 15 H; a thermal inertia of Figure 11. Serpent Bedform at Gusev crater. (a) Portion of 150 provides a better fit to the late afternoon data. Thus MI image 2M132842058FFL2000P2977 (sol 073) of the there is a phase lag in which measured temperatures cool disturbed bedform Serpent indicating a trimodal distribu- more slowly in the afternoon than modeled temperatures. tion of particle sizes of 1–2 mm (possibly bonded), 250– This complex behavior suggests that this surface is 500 mm, and <210 mm. (b) Plot of modeled subsurface retaining more heat in the afternoon than the thermal temperature at midnight of 1 mm granules (bedform armor), model predicts and that the processes operating at this site 300 mm sand grains (interior of bedform), and surface are not adequately represented in the thermal model. One temperature of 415 mm sand grains (Mini-TES–derived possible explanation for this behavior is a layer of particle size) at midnight. The circles at the ‘‘surface’’ are a bedrock close to the surface, such as that observed in scale representation of the size of the bedform armor the walls of Eagle and Endurance craters. This would relative to the subsurface temperature waveforms.

12 of 18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21

bedform, and this lower particle size may explain why bedforms do not develop in this material. [39] The thermal inertia modeled for bedforms within both Bonneville and Endurance craters (160–200) is lower than the average intracrater bedform thermal inertia derived from Viking and TES orbital data (270), but does fall within the range of bedform thermal inertias for intracrater deposits observed from orbit [e.g., Edgett and Christensen, 1991, 1994; Aben, 2003; Fenton et al., 2003]. In addition, Christensen [1983] noted that the average grain size of intracrater deposits often decreases with decreasing crater size; both Bonneville and Endurance are small relative to craters observed from orbit. Therefore the lower thermal inertia and grain size within these craters is expected. [40] Edgett and Christensen [1994] observed a correla- tion between intracrater deposit thermal inertias and the thermal inertia of the surrounding regolith from orbit, and suggested that there may be regional-scale controls on the particle size of intracrater material. At both landing sites, the thermal inertias measured for the soil and intracrater bedforms exhibit similar values between 100 and 200, corresponding to a particle diameter of 60 to 160 mm (silt to fine sand). Thus the thermal inertia of the intracrater bedforms often reflects the physical properties of the regional soil, and this observation suggests that the sediment for these bedforms may be locally derived. However, sand grains can be transported hundreds of km before they break into finer particles that go into suspension [Greeley and Kraft, 2001; Rogers and Christensen, 2003]. There- fore some of these materials may have originated outside the landing site regions as well. Laboratory measurements of particle-size mobility indicate that 100–150 mm is the Figure 12. Meridiani Plains soil. (a) Typical example most easily moved particle size on Mars [Iversen et al., of the Meridiani Plains soil (MI image 1976; Greeley et al., 1980]. The observation that intra- 1M133776216FFL08A6P2957, sol 063). (b) Plot of the crater bedforms commonly observed at these locations modeled diurnal temperature curve (using an opacity of have particle sizes overlapping this range suggests that 0.63 and Ls of 24.0°) and Mini-TES target temperatures; this material was transported into the crater by wind RMS error is 6.3 K for an inertia of 100 and 5.9 K for an potentially under current Martian atmospheric conditions, inertia of 150. but that this wind is currently unable to carry these particles back out onto the plains. This could be due to measured directly using MI images at the Meridiani plains. the wind dynamics inside the crater, such as reverse flow However this is the only locality that is problematic, and off the crater wall [Greeley et al., 1974], which does not particle sizes measured at other sites correspond well with allow the particles to be transported out of the crater. those derived from thermal inertia values. 3.2.2.2. Intracrater Deposits in Endurance Crater [38] The entire bedform located on the floor of Endurance 4. Implications crater has a modeled thermal inertia of 200, corresponding 4.1. Thermally Derived Albedo to 160 mm diameter particles (fine sand) (Figures 3 and 4). [41] The thermally derived albedo values are 17% This deposit also provides an opportunity to determine if smaller than the Pancam albedo values in most cases particle sizes between bedform crests and intercrest areas (Table 1). This lower albedo suggests that the surface is that fill the Mini-TES FOV can be differentiated. The absorbing more energy than the measured Pancam albedo temperatures vary by up to 5°K near noon across the indicates. This discrepancy was also observed in Viking bedform, but because of the large Mini-TES footprint it is orbital data [e.g., Kieffer et al., 1977; Palluconi and difficult to correlate temperature variations with bedform Kieffer, 1981; Hayashi et al., 1995]. The Viking IRTM morphology. These observed temperature differences are thermal model assumed an atmosphere whose contribution likely caused by a combination of a change in particle size to the downwelling radiation is equal to 2% of the maxi- from the bedform crest to trough and the differential heating mum solar isolation, and the thermally derived albedo caused by sloped surfaces. The aeolian material surrounding values are often smaller than the measured bolometric the intracrater bedform appears to be well sorted, and displays albedo [Kieffer et al., 1977; Palluconi and Kieffer, 1981]. no bedform morphology. The thermal inertia of this material Even with the incorporation of a more complex atmosphere, is 100, corresponding to particle diameters <45 mm (silt), including the effect of a dusty CO2 atmosphere and a which is lower than the calculated thermal inertias of the sensible heat exchange with the surface [Haberle and

13 of 18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21

338 nm broadband filter [Bell et al., 2004a, 2004b, 2006]) by 5–10% because on average the surface reflectivity is higher at that wavelength than in the shorter wavelengths closer to the peak of the Sun’s radiance spectrum. If the surface reflectance is not Lambertian, then the amount of reflectance changes with incidence and emission angle, and the measured albedo varies depending on the viewing geometry and local time of the observation [Hapke, 1993, pp. 190–192]. Therefore the albedo calculated from individual Pancam images using the 750 nm band may not best represent the true albedo of the surface. This uncertainty is affecting the relative difference between these two values, and is also contributing to errors in the amount of energy the model assumes is absorbed into the subsurface, which is compensated by the thermally derived albedo.

4.2. Comparison With TES and THEMIS Orbital Data [44] The Thermal Emission Spectrometer (TES) [Christensen et al., 1992] and the Thermal Emission Imaging System (THEMIS) [Christensen et al., 2004a] Figure 13. THEMIS thermal inertia of the Gusev landing obtain orbital thermophysical information of the Martian site region. THEMIS-derived thermal inertia image overlaid surface at a spatial resolution of 3km6 km per onto a THEMIS visible image of the traverse from pixel and 100 m per pixel respectively. The Mini-TES– Columbia Memorial Station (CMS) to Bonneville crater to derived thermal inertias provide an opportunity to validate the Columbia Hills at the Gusev landing site region. these orbital data sets, and offer insight regarding the Numbers indicate site positions along the traverse. Site interpretation of orbital thermal inertia data. position along the Gusev traverse was typically incremented [45] The Spirit landing site at Gusev has an average TES after each drive segment and was chosen because site is thermal inertia of 290 [Christensen et al., 2004b], more representative of the distance traveled than sol suggesting that the surface is dominated by duricrust to numbers. cemented soil, noncohesive coarse sand, or a combination. This value is consistent with the variety of surface types observed by the Spirit rover. The comparison of THEMIS- Jakosky, 1991], the thermally derived albedo was often derived thermal inertia [Fergason and Christensen, 2003] lower than the measured albedo values for bright surfaces with Mini-TES thermal inertia along the traverse from the and higher than the measured value for dark surfaces CMS to Bonneville crater and the Columbia Hills is shown [Hayashi et al., 1995]. in Figure 8 and Figure 13. Although the difference in [42] Factors likely contributing to the discrepancy resolution between the two data sets presents a challenge between thermally derived albedo values and measured when comparing the resulting thermal inertia directly, bolometric albedo include thermal model uncertainties in inspection of THEMIS-derived thermal inertia and Mini- the amount of absorbed and reflected energy (albedo) TES thermal inertia shows that the general trends match and the amount of downwelling or backscatter radiance between data sets. The thermal inertia increases along the from the atmosphere or nearby objects. Model parameters traverse from CMS (THEMIS: 280 ± 40; Mini-TES: 175 ± controlling atmospheric effects, such as downwelling 20) to Bonneville crater (THEMIS: 330 ± 50; Mini-TES: radiance and the visible/9-mm extinction opacity ratio 380 ± 45) in both data sets, and is likely due to more may be inaccurately defined because of a lack of under- rocky material being present as the crater ejecta material standing concerning atmospheric behavior. These uncer- is traversed [Christensen et al., 2004a; Golombek et al., tainties will affect the modeled diurnal temperature curves 2005]. The thermal inertia is more variable along the and thermally derived albedo that best fit the data. The traverse from Bonneville crater to the Columbia Hills, visible/9-mm extinction opacity ratio may range from 2.0 but increasing and decreasing thermal inertia patterns are to 2.5 for average atmospheric conditions [Clancy et al., similar and the average value of each data set along 1995]. Uncertainties in this ratio results in 10% uncer- this traverse differs by 35 units. THEMIS thermal tainty (0.5–1 K during the day) and this error alone is not inertias near the Colombia Hills (390 ± 45) are larger than sufficient to cause the 2.5 K daytime temperature differ- Mini-TES (275 ± 35) due to the incorporation of the ence observed for surfaces with a 0.03 difference in higher-inertia hill material in the THEMIS pixel, whereas albedo (average difference between thermally derived the Mini-TES is sampling surface material that excludes albedo and Pancam albedo). rocky terrain adjacent to the rover (within 1m). [43] The Pancam albedo assumes a Lambertian surface Generally the thermal inertia measured from the surface reflectance [Bell et al., 2006], which is not precise for most is lower than values observed from orbit. These lower natural surfaces [e.g., Hapke, 1993, pp. 190–191]. Using surface thermal inertia values may indicate a data bias, as the 750 nm band to estimate R* typically over-estimates the Mini-TES observations were taken directly in front of the bolometric albedo (calculated using the Pancam 739 ± the rover, and during the rover traverse, obstacles, such as

14 of 18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21

Table 2. Comparison of Mini-TES–Derived Thermal Inertia, Thermally Derived Particle Sizes, and Particle Sizes Measured Directly Using MI Imagesa Mini-TES–Derived Thermal Inertia, Mini-TES–Derived MI-Measured Feature J/m2 Ks1/2 Particle Size Particle Size Laguna Hollow (Gusev) 150 45 mm <210 mm Gusev Plains Soil 200 160 mm fine grains or agglomerates of grains at the margin of resolution (210 mm) Serpent/Saber Bedform (Gusev) 250 415 mm trimodal: 1–2 mm, 250 – 500 mm, <210 mm Meridiani Plains Soil 100 to 150 45 mm bimodal: 0.6–6 mm set in very fine sand less <125 mm aMI particle sizes from the Gusev landing site region are from Herkenhoff et al. [2004a]. MI particle sizes from the Meridiani landing site region are from Herkenhoff et al. [2004b]. bedforms or large rocks, were avoided. However, we now and Christensen, 1997a] to estimate an effective particle can document that there is variability and mixing of size of a surface observed remotely. particle sizes along the traverse at Gusev crater that is [48] Coincident MI images and thermal inertia informa- not resolved in the THEMIS thermal inertia. tion were collected at four locations: the material filling the [46] The Meridiani plains provide a favorable opportunity hollows at Gusev, the Gusev soil, the Serpent/Saber bed- to validate orbital data because the plains are uniform over forms at Gusev, and the Meridiani Plains material (Table 2). several TES and THEMIS pixels. The Opportunity landing The thermally derived particle sizes are often near or below site at Meridiani has a TES thermal inertia of 200 and a the limit of MI resolution (100 mm, which corresponds to a THEMIS thermal inertia of 190 ± 30, implying a particle thermal inertia of 180), but many of these observations are size of fine sand [Golombek et al., 2005]. Orbitally derived in agreement within the uncertainties of the thermal inertia thermal inertia values are also consistent with particle sizes calculations. In the hollows at Gusev crater, MI images estimated from Pancam and MI imagery. The Mini-TES– revealed that the majority of the particles are at or below the derived thermal inertias, however, give values of 100 to 150 100 mm limit of the MI camera. Particle sizes determined (Figure 12), which are lower than the orbital values, and with Mini-TES (thermal inertia of 150) suggest 45 mm imply particle sizes (10 to 45 mm) that are lower than diameter particles (silt), which is also below the resolution expected for the observed bedforms. There are several limit of MI. The soil along the Gusev plains is composed of potential explanations for this difference. The thermal very fine grains or conglomerates of grains at the margin of model assumes a Lambertian visible scattering surface, MI resolution [Herkenhoff et al., 2004a]. Mini-TES derives where light energy is reflected equally in all directions. an average particle size of Gusev soil outside the hollows Results of a photometry experiment at this same location (thermal inertia of 200) to be 160 mm, corresponding to suggests a Minnaert surface, which also takes into account fine sand, and is in agreement with MI. the viewing geometry, is a more accurate representation [49] At the Serpent bedform in Gusev the particle size [Seelos et al., 2005]. The photometric properties of the distribution measured in MI images revealed a trimodal surface affects the amount of absorbed energy assumed by distribution of particles 1–2 mm, 250–500 mm, and below the thermal model, and thus this parameter may not be well 210 mm[Herkenhoff et al., 2004a]. The Mini-TES–derived represented. However, it has been demonstrated that view- thermal inertia of Saber (250), a similar bedform near ing sand-sized surfaces at different phase functions does not Serpent, suggests a particle size of 415 mm, which is also typically change the surface temperature behavior [Jakosky consistent with grains measured by the MI. Serpent and et al., 1990], and thus a difference in the actual surface Saber probably have a layer of coarse particles that form an photometry relative to what is represented in the model outer armor over finer-grained material [Greeley et al., may not be significant. An additional possibility is that 2004]. A single grain thickness of armor (1–2 mm particles) the model does not adequately account for all the surface will increase the thermal inertia, but in this layered case the and subsurface physical properties that are occurring at material is still dominated by the finer-grained substrate this location because the fit of modeled temperatures to (Figure 11). Thus particle sizes derived from Mini-TES those determined by Mini-TES at the Meridiani Plains has thermal inertia values and directly measured particle size a higher RMS residual relative to other MER sites. Since distributions using MI images do compare favorably within TES and THEMIS both use a single-point measurement to the uncertainties in thermal inertia calculations. calculate thermal inertia values, this result would not be [50] On the Meridiani plains, modeled temperatures apparent from orbital data. did not well represent the Mini-TES target temperatures. The best-fit Mini-TES thermal inertia values of 100 to 4.3. Particle Size Comparisons 150 indicate an average particle diameter of <45 mm (silt), [47] Orbital thermal inertia data are commonly used to which is smaller than particles capable of being measured determine an effective particle size, and thus thermal inertia using MI images. A bimodal particle size distribution of influences the interpretation of landforms and morphology 0.6–6 mm particles (spherules) set in a very fine sand observed from orbit. One objective of this study is to derive less than 125 mm is observed with MI [Herkenhoff et al., particle sizes from thermal inertia values and compare the 2004b]. Since the spherules are only a single-grain-thick results to particle sizes measured directly using MI images. layer, rather than an intimate mixture, they do not This will help to validate the use of the laboratory-derived dominate the thermal inertia of this material. However, a relationship between conductivity and particle size [e.g., thermal inertia of 190 corresponds to the MI observed Wechsler and Glaser, 1965; Wechsler et al., 1972; Presley particle size of 125 mm, so there is a discrepancy. The

15 of 18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21 grains could be vesicular, which would reduce their [53] It has been proposed that the hematite spherules are conductivity and modeled thermal inertia relative to the remnant lag deposit from a material that has been removed particle size observed with MI. This is unlikely, however, by erosion [Squyres et al., 2004b; Soderblom et al., 2004]. since the MI-measured particle sizes correspond well with The lack of craters and ejecta present on the Meridiani orbital thermal inertia data. The cause of this discrepancy plains also suggests that this surface has undergone exten- is not understood. However, the results from the MER sive erosion in its history, and is in agreement with landing sites indicate that in most cases effective particle conclusions from orbital data [e.g., Christensen and Ruff, sizes derived from thermal inertia agree with directly 2004; Lane et al., 2003; Hynek et al., 2002]. Further measured size distributions, and that relationships between evidence for aeolian activity is found on the plains surface conductivity and particle size observed by the MER where aeolian bedforms are pervasive throughout the region rovers are similar to that deduced in the laboratory. [Sullivan et al., 2005]. Unlike Gusev, these bedforms, and those found in craters, are dark, suggesting little dust has 4.4. Geology and Implications been deposited and implying that they may be currently [51] The thermal inertia along the Gusev plains traverse active or have been active in the recent past. varies by less than 90 units, and these changes can typically be attributed to the presence of rock fragments or aeolian material in the Mini-TES FOV. Pancam images taken of the 5. Conclusions same location as the Mini-TES observations show that the [54] The Mini-TES instrument provides the first oppor- variations in thermal inertia correspond with the amount of tunity to measure surface temperatures in-situ, derive ther- rock fragments in the area, where the number and size of mal inertia values, and relate these observations to orbital rock fragments increases in higher thermal inertia regions results. Derived thermal inertia along the Gusev traverse [Ward et al., 2005]. An increase in thermal inertia and and of bedforms at both sites are consistent with orbital rock abundance is also associated with crater ejecta [e.g., values, and significantly improve our confidence in the Golombek et al., 2005, 2006]. These observations suggest accuracy and interpretation of orbital data. In addition, that at meter scales, the thermal inertia is controlled by comparisons between thermally derived particle sizes and the abundance of drift material, rocks, and rock fragments, particle sizes measured directly using MI images imply that whose local distribution is controlled by local processes laboratory-derived relationships between conductivity and that redistribute grains and rocky material. In contrast, particle size are applicable to the surface of Mars. From this Christensen [1982] examined the relationship between work, we conclude the following: rock abundance and the fine component thermal inertia [55] 1. Bedforms in the floor of Bonneville crater at with global thermal inertia patterns and concluded that at Gusev crater and Endurance crater at Meridiani Planum global scales the variations in the fine component (material have a thermal inertia of 160 to 200 and 200, respectively, less than a few mm) thermal inertia, rather than the corresponding to particle diameters of 160 mm (fine sand). abundance of rocks, produce much of the variation in Similar thermal inertias suggest that they formed under the thermal inertia. These observations imply that different same atmospheric conditions, and that particles are able to processes are affecting the physical surface characteristics be mobile in the current climate. Bright albedo bedforms at at different scales. Rocky material from crater ejecta, Bonneville crater imply that these features are mantled in presence of aeolian bedforms, and rock fragment concen- dust and have not been recently active. This is in contrast trations control the thermal inertia at meter to tens of with bedforms observed in Endurance crater, whose dark meter scales. Regional processes, such as the deposition albedo suggests recent aeolian activity. of airfall dust, deflation or deposition of wind-blown sand, [56] 2. In Laguna Hollow at Gusev crater, the modeled or large-scale volcanic processes, control the thermal inertia thermal inertia is 150, which corresponds to 45 mm diameter at km to tens of km scales. particles (silt), and is some of the finest-grained material [52] The rock and bedrock material on the plains and observed at either landing site region. These small craters in the Columbia hills have different physical characteristics have been filled with a combination of fine material and that imply that these regions have a different geologic air-fall dust. history. The rocks on the Gusev plains are dense basalt [57] 3. At Gusev crater, the rock target Beacon has a with a thermal inertia of 1200. At the lower Columbia thermal inertia of 1200, which is close to that expected for a Hills, however, the average thermal inertia is 620 (max- basaltic rock this size. In contrast, the rock and bedrock at imum is 1100) and suggests that Columbia Hills rocks are the lower Columbia Hills have an average thermal inertia of more weathered and friable than those found on the plains. 620 (maximum value of 1100). These different thermal The amount of energy per unit volume necessary to grind inertia values imply that thermal inertia can differentiate these rocks with the RAT instrument was also less than between rocks of differing degrees of density or porosity. rocks on the plains (Gorevan et al., manuscript in prepa- The plains rocks may be derived from a basaltic lava ration, 2006), supporting this interpretation. The plains flow, whereas the lower Colombia Hills rocks have the rocks may be the result of a basaltic lava flow, and may physical characteristics of a volcaniclastic sediment, ash have been either exhumed from the subsurface or broken flow deposit, or impact debris. into unconsolidated soil by impact events. The lower [58] 4. Particle sizes derived from thermal inertia corre- Colombia Hills rock has the physical properties of a spond to directly measured particle sizes using MI imagery volcaniclastic sediment, ash flow deposit, or impact debris, in material filling hollows, the Gusev soil, and the Serpent/ implying volcanic activity different from the effusion of Saber bedforms. These results suggest that techniques basalt as the source for material in this region. derived from laboratory measurement do provide a rea-

16 of 18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21 sonable derivation of particle sizes from thermal inertia Christensen, P. R. (1983), Eolian intracrater deposits on Mars: Physical values. properties and global distribution, Icarus, 56, 496–518. Christensen, P. R. (1986), The spatial distribution of rocks on Mars, [59] 5. At Gusev crater, the thermal inertia measured from Icarus, 68, 217–238. the surface corresponds to trends observed from orbital data. Christensen, P. R., and S. W. Ruff (2004), Formation of the hematite- There is variability and mixing of particle sizes on the bearing unit in Meridiani Planum: Evidence for deposition in standing water, J. Geophys. Res., 109, E08003, doi:10.1029/2003JE002233. surface that is not resolved in the orbital thermal inertia data Christensen, P. R., et al. (1992), Thermal Emission Spectrometer experi- due to meter-scale processes that are not identifiable at ment: Mars Observer Mission, J. Geophys. Res., 97(E5), 7719–7734. larger scales. On the Meridiani Plains, orbital data are Christensen, P. R., et al. (2003), Miniature Thermal Emission Spectrometer for the Mars Exploration Rovers, J. Geophys. Res., 108(E12), 8064, consistent with visible observations in Pancam and MI doi:10.1029/2003JE002117. imagery, but correspond poorly with Mini-TES–derived Christensen, P. R., et al. (2004a), The Thermal Emission Imaging System thermal inertias of 100 to 150. (THEMIS) for the Mars 2001 Odyssey mission, Space Sci. Rev., 110, 85–130. [60] 6. The diurnal temperature behavior of the Meridiani Christensen, P. R., et al. (2004b), Initial results from the Mini-TES plains is not well characterized by the thermal model used in experiment in Gusev Crater from the Spirit Rover, Science, this work. In addition, the Mini-TES–derived thermal 305(5685), 837–842. inertias are also lower than that observed with TES and Clancy, R. T., S. W. Lee, G. R. Gladstone, W. W. McMillan, and T. Rousch (1995), A new model for Mars atmospheric dust based upon analysis of THEMIS from orbit. The cause of this discrepancy is ultraviolet through infrared observations from Mariner 9, Viking, and unknown at this time, but only occurs on surfaces where Phobos, J. Geophys. Res., 100(E3), 5251–5263. hematite spherules are present. Edgett, K. S., and P. R. Christensen (1991), The particle size of Martian aeolian dunes, J. Geophys. Res., 96(E5), 22,765–22,776. [61] 7. In most cases, the thermally derived albedo is Edgett, K. S., and P. R. Christensen (1994), Mars aeolian sand: Regional 17% lower than the Pancam estimated bolometric albedo. variations among dark-hued crater floor features, J. Geophys. Res., This difference suggests that the surface is absorbing more 99(E1), 1997–2018. energy than that predicted by the models that incorporate Fenton, L. K., J. L. Bandfield, and A. W. Ward (2003), Aeolian processes in Proctor Crater on Mars: Sedimentary history as analyzed from multiple the measured albedo and atmospheric properties. The pri- data sets, J. Geophys. Res., 108(E12), 5129, doi:10.1029/2002JE002015. mary causes of this discrepancy are uncertainties in the Fergason, R. L., and P. R. Christensen (2003), Thermal inertia using downwelling radiance, the visible/9-mm extinction opacity THEMIS infrared data, Lunar Planet. Sci. [CD-ROM], XXXIV, Abstract 1785. ratio, and the photometric properties of the surface. Fountain, J. A., and E. A. West (1970), Thermal conductivity of particulate basalt as a function of density in simulated lunar and Martian environ- [62] Acknowledgments. We wish to thank Mark Lemmon for pro- ments, J. Geophys. Res., 75(20), 4063–4069. viding Pancam opacity data to the MER Athena Science Team. Frank Golombek, M. P., et al. (2005), Assessment of Mars Exploration Rover Palluconi and James Zimbelman provided reviews that significantly im- landing site predictions, Nature, 436, doi:10.1038/nature03600. proved the quality of this manuscript. Tim Glotch, Deanne Rogers, Steve Golombek, M. P., et al. (2006), Geology of the Gusev cratered plains from Ruff, and Mike Wyatt also contributed discussions and suggestions that the Spirit rover transverse, J. Geophys. Res., 111, E02S07, doi:10.1029/ improved this work. In addition, Saadat Anwar, Noel Gorelick, and Dale 2005JE002503. Noss provided programming assistance, Mike Wyatt helped with the Grant, J. A., et al. (2004), Surficial deposits at Gusev Crater along Spirit THEMIS overlay figure, and Jascha Sohl-Dickstein helped with calculating Rover traverses, Science, 305, 807–810. Pancam estimated bolometric albedo. We also wish to thank the Mini-TES, Greeley, R., and M. D. Kraft (2001), Survivability of aggregate sands on Pancam, and Microscopic Imager Mission Operations staff for collecting Mars, Lunar Planet. Sci. [CD-ROM], XXXII, Abstract 1839. and calibrating the data products used in this study. This work was Greeley, R., J. D. Iversen, J. B. Pollack, N. Udovich, and B. White (1974), performed for the Jet Propulsion Laboratory, California Institute of Wind tunnel studies of Martian aeolian processes, Proc. R. Soc. London, Technology, under a contract with the National Aeronautics and Space Ser. A, 341, 331–360. Administration. Greeley, R., R. Leach, B. White, J. Iversen, and J. Pollack (1980), Thresh- old windspeeds for sand on Mars: Wind tunnel simulations, Geophys. Res. Lett., 7(2), 121–124. References Greeley, R., et al. (2004), Wind-related processes detected by the Spirit Aben, L. K. (2003), Compositional and thermophysical analysis of Mar- Rover at Gusev Crater, Science, 305(5685), 810–821. tian aeolian dunes, M.S. thesis, 117 pp., Ariz. State Univ., Tempe, Dec. Greeley, R., et al. (2006), Gusev crater: Wind-related features and processes Arvidson, R. E., et al. (2004a), Localization and physical properties observed by the Mars Exploration Rover Spirit, J. Geophys. Res., 111, experiments conducted by Spirit at Gusev Crater, Science, 305(5685), E02S09, doi:10.1029/2005JE002491. 821–824. Haberle, R. M., and B. M. Jakosky (1991), Atmospheric effects on the Arvidson, R. E., et al. (2004b), Localization and physical properties experi- remote determination of thermal inertia on Mars, Icarus, 90, 187– ments conducted by Opportunity at Meridiani Planum, Science, 204. 306(5702), 1730–1733. Hapke, B. (1993), Theory of Reflectance and Emittance Spectroscopy, 455 Arvidson, R. V., et al. (2006), Overview of the Spirit Mars Exploration pp., Cambridge Univ. Press, New York. Rover Mission to Gusev Crater: Landing site to Backstay Rock in the Hayashi, J. N., B. M. Jakosky, and R. M. Haberle (1995), Atmospheric Columbia Hills, J. Geophys. Res., 111, E02S01, doi:10.1029/ effects on the mapping of Martian thermal inertia and thermally derived 2005JE002499. albedo, J. Geophys. Res., 100(E3), 5277–5284. Bell, J. F., III, et al. (2003), Mars Exploration Rover Athena Panoramic Herkenhoff, K. E., et al. (2003), Athena Microscopic Imager investigation, Camera (Pancam) investigation, J. Geophys. Res., 108(E12), 8063, J. Geophys. Res., 108(E12), 8065, doi:10.1029/2003JE002076. doi:10.1029/2003JE002070. Herkenhoff, K. E., et al. (2004a), Textures of the soils and rocks at Bell, J. F., III, et al. (2004a), Pancam multispectral imaging results from the Gusev Crater from Spirit’s Microscopic Imager, Science, 305(5685), Spirit Rover at Gusev Crater, Science, 305(5685), 800–806. 824–826. Bell, J. F., III, et al. (2004b), Pancam multispectral imaging results from Herkenhoff, K. E., et al. (2004b), Evidence from Opportunity’s Micro- the Opportunity Rover at Meridiani Planum, Science, 306(5702), scopic Imager for water on Meridiani Planum, Science, 306(5702), 1703–1709. 1727–1730. Bell, J. F., III, J. Joseph, J. N. Sohl-Dickstein, H. M. Arneson, M. J. Hynek, B. M., R. E. Arvidson, and R. J. Phillips (2002), Geologic setting Johnson, M. T. Lemmon, and D. Savransky (2006), In-flight calibration and origin of Terra Meridiani hematite deposit on Mars, J. Geophys. Res., and performance of the Mars Exploration Rover Panoramic Camera 107(E10), 5088, doi:10.1029/2002JE001891. (Pancam) instruments, J. Geophys. Res., 111, E02S03, doi:10.1029/ Iversen, J. D., J. B. Pollack, R. Greeley, and B. R. White (1976), Saltation 2005JE002444. threshold on Mars: The effect of interparticle force, surface roughness, Christensen, P. R. (1982), Martian dust mantling and surface composition: and low atmospheric density, Icarus, 29, 381–393. Interpretation of thermophysical properties, J. Geophys. Res., 87(B12), Jakosky, B. M. (1979), The effects of nonideal surfaces on the derived 9985–9998. thermal properties of Mars, J. Geophys. Res., 84(B14), 8252–8262.

17 of 18 E02S21 FERGASON ET AL.: PHYSICAL PROPERTIES OF THE MER LANDING SITES E02S21

Jakosky, B. M. (1986), On the thermal properties of Martian fines, Icarus, Rogers, D., and P. R. Christensen (2003), Age relationship of basaltic and 66, 117–124. andesitic surface composition on Mars: Analysis of high-resolution TES Jakosky, B. M., G. W. Finiol, and B. G. Henderson (1990), Directional observations in the northern hemisphere, J. Geophys. Res., 108(E4), variations in thermal emission from geologic surfaces, Geophys. Res. 5030, doi:10.1029/2002JE001913. Lett., 17(7), 985–988. Ruff, S. W., and P. R. Christensen (2002), Bright and dark regions on Mars: Jakosky, B. M., M. T. Mellon, H. H. Kieffer, P. R. Christensen, E. S. Particle size and mineralogical characteristics based on Thermal Emission Varnes, and S. W. Lee (2000), The thermal inertia of Mars from the Spectrometer data, J. Geophys. Res., 107(E12), 5127, doi:10.1029/ Mars Global Surveyor Thermal Emission Spectrometer, J. Geophys. 2001JE001580. Res., 105, 9643–9652. Seelos, F. P., R. E. Arvidson, E. A. Guinness, M. J. Wolff, and the Athena Kahle, A. B. (1980), Surface thermal properties, in Remote Sensing in Science Team (2005), Radiative transfer photometric analysis at the Mars Geology, edited by B. S. Siegal and A. R. Gillespie, pp. 257–273, John Exploration Rover landing sites, Lunar Planet. Sci. [CD-ROM], XXVI, Wiley, Hoboken, N. J. Abstract 2054. Kieffer, H. H., S. C. Chase Jr., E. Miner, G. Mu¨nch, and G. Neugebauer Smith, D. E., and M. T. Zuber (1998), The relationship between MOLA (1973), Preliminary report on infrared radiometric measurements from northern hemisphere topography and the 6.1-mbar atmospheric pressure the Mariner 9 spacecraft, J. Geophys. Res., 78(20), 4291–4312. surface of Mars, Geophys. Res. Lett., 25(24), 4397–4400. Kieffer, H. H., T. Z. Martin, A. R. Peterfreund, B. M. Jakosky, E. D. Minert, Smith, D. E., et al. (2001), Mars Orbiter Laser Altimeter: Experiment and F. D. Palluconi (1977), Thermal and albedo mapping of Mars during summary after the first year of global mapping of Mars, J. Geophys. the Viking primary mission, J. Geophys. Res., 82(28), 4249–4291. Res., 106(E10), 23,689–23,722. Kieffer, H. H., B. M. Jakosky, and C. W. Snyder (1992), The planet Mars: Soderblom, L. A., et al. (2004), Soils of Eagle Crater and Meridiani Planum From antiquity to the present, in Mars, edited by H. H. Kieffer et al., at the Opportunity Rover landing site, Science, 306(5702), 1723–1726. pp. 1–33, Univ. of Ariz. Press, Tucson. Squyres, S. W., et al. (2004a), The Spirit Rover’s Athena science investiga- Lane, M. D., P. R. Christensen, and W. K. Hartmann (2003), Utilization of tion at Gusev Crater, Mars, Science, 305(5685), 794–799. the THEMIS visible and infrared imaging data for crater population Squyres, S. W., et al. (2004b), The Opportunity Rover’s Athena science studies of the Meridiani Planum landing site, Geophys. Res. Lett., investigation at Meridiani Planum, Mars, Science, 306(5702), 1698– 30(14), 1770, doi:10.1029/2003GL017183. 1703. Lemmon, M. T., et al. (2004), Atmospheric imaging results from the Mars Sullivan, R., et al. (2005), Aeolian processes at the Mars Exploration Rover Exploration Rovers: Spirit and Opportunity, Science, 306(5702), 1753– Meridiani Planum landing site, Nature, 436, doi:10.1038/nature03641. 1756. Toon, O. B., J. B. Pollack, and C. Sagan (1977), Physical properties of the Mellon, M. T., B. M. Jakosky, H. H. Kieffer, and P. R. Christensen (2000), particles comprising the Martian dust storm of 1971–1972, Icarus, 3, High-resolution thermal inertia mapping from the Mars Global Surveyor 633–696. Thermal Emission Spectrometer, Icarus, 148, 437–455. Wang, A., et al. (2006), Sulfate deposition in subsurface regolith in Gusev Moersch, J. E., et al. (2005), Comparison of orbital infrared observations Crater, Mars, J. Geophys. Res., doi:10.1029/2005JE002513, in press. and surface measurements by the Mars Exploration Rover Spirit at Gusev Ward, J. G., R. E. Arvidson, and M. Golombek (2005), The size-frequency Crater, Lunar Planet. Sci. [CD-ROM], XXVI, Abstract 2020. and areal distribution of rock clasts at the Spirit landing site, Gusev crater, Neugebauer, G., G. Mu¨nch, H. Kieffer, S. C. Chase Jr., and E. Miner Mars, Geophys. Res. Lett., 32, L11203, doi:10.1029/2005GL022705. (1971), Mariner 1969 infrared radiometer results: Temperatures and Wechsler, A. E., and P. E. Glaser (1965), Pressure effects on postulated thermal properties of the Martian surface, Astron. J., 76(8), 719– lunar materials, Icarus, 4, 335–352. 728. Wechsler, A. E., P. E. Glaser, and J. A. Fountain (1972), Thermal properties Palluconi, F. D., and H. H. Kieffer (1981), Thermal inertia mapping of Mars of granulated materials, in Thermal Characteristics of the Moon, edited from 60°Sto60°N, Icarus, 45, 415–426. by J. W. Lucas, Progr. Astron. Aeronaut., 28, 215–241. Pollack, J. B., D. S. Colburn, F. M. Flasar, R. Kahn, C. E. Carlston, and Zimbelman, J. R. (1986), The role of porosity in thermal inertia variations D. Pidek (1979), Properties and effects of dust particles suspended in on basaltic lavas, Icarus, 68, 366–369. the Martian atmosphere, J. Geophys. Res., 84(B6), 2929–2945. Presley, M. A., and P. R. Christensen (1997a), Thermal conductivity measurements of particulate materials: 2. Results, J. Geophys. Res., J. F. Bell III, Department of Astronomy, Space Science Building, 102(E3), 6551–6566. Cornell University, Ithaca, NY 14853, USA. Presley, M. A., and P. R. Christensen (1997b), The effect of bulk density P. R. Christensen and R. L. Fergason, Department of Geological and particle size sorting on the thermal conductivity of particulate mate- Sciences, Arizona State University, Box 876305, Tempe, AZ 85287-6305, rials under Martian atmospheric pressures, J. Geophys. Res., 102(E4), USA. ([email protected]) 9221–9229. M. P. Golombek, Jet Propulsion Laboratory, California Institute of Reid, R. J., et al. (1999), Imager for Mars Pathfinder (IMP) image calibra- Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, USA. tion, J. Geophys. Res., 104(E4), 8907–8925. K. E. Herkenhoff, Astrogeology Team, U.S. Geological Survey, 2255 Robertson, E. C., and D. L. Peck (1974), Thermal conductivity of vesicular North Gemini Drive, Flagstaff, AZ 86001, USA. basalt from Hawaii, J. Geophys. Res., 79(32), 4875–4888. H. H. Kieffer, Celestial Reasonings, Carson City, NV 89703, USA.

18 of 18